Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Lecture 5
Quasi-Static Analyses
L5.2
Overview
Introduction Quasi-Static Simulations Using Explicit Dynamics Loading Rates Energy Balance in Quasi-Static Analyses Mass Scaling Viscous Pressure Summary
Introduction
L5.4
Introduction
Challenging nonlinear quasi-static problems often involve: Very complex contact conditions Very large deformations Mesh distortion possible Applications: Metal forming simulations: Bulk forming (drawing, rolling, extrusion, upsetting, etc.) Sheet metal forming (stretching, drawing)
Rolling of a symmetric I-section
Video Clip
L5.5
Introduction
Applications (contd): Quasi-static collapse analyses Example: Collapse of a curved beam:
undeformed shape
Quasi-static loading of flexible rubber components (seals, bushings, etc.) Example: Compression of a rubber gasket
L5.6
Introduction
ABAQUS offers two solvers: Implicit solver (ABAQUS/Standard) Solves for true static equilibrium. Explicit solver (ABAQUS/Explicit) Solves for true dynamic equilibrium. At first glance it appears the implicit solver would be the appropriate choice for modeling highly nonlinear static problems. However, explicit solvers are more efficient for this class of problems. This is especially true for three-dimensional problems involving contact and very large deformations.
L5.7
Introduction
Example: Simulation of a deep drawing process used to form an oil pan The pan is formed by displacing the punch downward while holding the die and blank holder fixed. The blank is modeled with shell elements; the tools are assumed rigid. Analysis performed with both implicit (ABAQUS/Standard) and explicit (ABAQUS/Explicit) solvers.
punch die blank
blank holder
L5.8
Introduction
The final deformed configuration is shown at right. Near the end of the punch stroke, the blank pulls through the blank holder and begins to wrinkle. The ABAQUS/Standard job was about 20 times more expensive than the ABAQUS/Explicit job (CPU cost). ABAQUS/Standard fails to converge at the point where the blank begins to wrinkle.
L5.10
L5.11
L5.12
Loading Rates
L5.14
Loading Rates
The dominant response of a quasi-static analysis will be the first structural mode. The frequency of this mode is used to estimate the impact velocity. Estimate the first natural frequency of the model. Calculate the corresponding time period (T) using the first natural frequency of the model. Estimate the global deflection (D) in the impact direction of the model Calculate the impact velocity (V) by using the formula V=D/T A general recommendation is to limit the impact velocity to less than 1% of the wave speed of the material Typical wave speed in metals is 5000 m/sec.
L5.15
Loading Rates
Suggested approach Run a series of simulations in the order from the fastest load rate to the slowest, since the analysis time is greater for slower load rates. Examine the results (deformed shapes, stresses, strains, energies) to get an understanding for the effects of varying the model. For example, excessive tool speeds in explicit sheet metal forming simulations tend to suppress wrinkling and to promote unrealistic localized stretching. Excessive tool speeds in explicit bulk forming simulations cause jettinghydrodynamic-type response. Excessive loading rates in a quasi-static collapse analysis can result in a steep initial slope of the load versus displacement curve inertial effects cause increased (non-structural) resistance to initial deformation localized buckling near the applied load
Copyright 2005 ABAQUS, Inc.
L5.16
Loading Rates
Jetting Consider the following bulk forming process (180 section of an axisymmetric model). When the tool speed is too large, highly localized deformation develops (jetting).
Video Clip
jetting
L5.17
Loading Rates
Example: Door beam intrusion test A simple model of a standard door beam intrusion test for an automobile door is shown. The circular beam is fixed at each end, and the beam is deformed by a rigid cylinder. The actual test is quasistatic.
L5.18
Loading Rates
Example (contd) : Door beam intrusion test An initial analysis is run with an extremely high impact velocity, 400 m/sec. The final deformed shape indicates a dynamic event rather than a static one: There is highly localized deformation No structural response by the beam
V
L5.19
Loading Rates
Example (contd) : Door beam intrusion test The dominant response in a static test will be in the first structural mode of the beam. The frequency of this mode is used to estimate the impact velocity. The frequency of the first mode is approximately 250 Hz. This rate corresponds to a period of 4 milliseconds. Using a velocity of 25 m/sec, the cylinder will be pushed into the beam 0.1 m in 4 milliseconds.
Video Clip
0.1 m
Velocity 25 m/s
L5.20
Loading Rates
Why is the velocity 25 m/sec appropriate? The frequency (f) of the first mode is approximately 250 Hz. This corresponds to a period t=0.004 seconds. During this period, the rigid cylinder is pushed into the beam d=0.1 m. Thus, the velocity v is estimated to be v = d / t = 0.1/0.004=25 m/sec. Recall, the wave speed of metals is about 5000 m/sec, so the impact velocity 25 m/sec is about 0.5% of the wave speed. The impact velocity should be limited to less than 1% of the wave speed of the material. A more accurate solution could be obtained by ramping up the velocity smoothly from zero over the analysis step.
L5.21
Loading Rates
Use the SMOOTH STEP amplitude curve A quasi-static solution is also promoted by applying loads gradually: Instantaneous loading may induce the propagation of a stress wave through the model, producing undesired results. By default, ABAQUS/Explicit loads applied immediately and remain constant throughout the step. Constant velocity boundary conditions also result in a sudden impact load onto a deformable body. Ramping up the loading gradually from zero minimizes these adverse effects. Ramping down the loading to zero is also recommended for the same reasons.
L5.22
Loading Rates
Syntax example (single step):
*AMPLITUDE, NAME=SMOOTH, DEFINITION=SMOOTH STEP 0.0, 0.0, 1.e-5, 1.0 *BOUNDARY, TYPE=DISPLACEMENT, AMP=SMOOTH 12, 2, 2, 2.5 A fifth-order polynomial transition between two amplitude values is created such that the first and second time derivatives are zero at the beginning and the end of the transition.
L5.23
Loading Rates
Syntax example (multiple steps):
*AMPLITUDE, NAME=ex2, DEFINITION=SMOOTH STEP, TIME=TOTAL TIME 0.0, 0.1, 0.1, 0.1, 0.2, 0.3, 0.3, 0.5 0.4, 0.5, 0.5, 0.2, 0.8, 0.2
L5.25
Consider a pull test applied to a uniaxial tensile specimen. If the physical test is quasi-static, the work applied by the external forces in stretching the specimen equals the internal energy in the specimen.
Uniaxial pull test
L5.26
L5.27
L5.28
L5.29
L5.30
STH
+9.355e-04 +8.500e-04 +8.250e-04 +8.000e-04 +7.750e-04 +7.500e-04 +7.250e-04 +7.000e-04 +2.328e-04
Vpunch = 30 m/s
Video Clip
L5.31
Mass Scaling
L5.33
Mass Scaling
Introduction When the loading rates is increased to efficiently model a quasi-static problem, the material strain rates calculated in the simulation are artificially high by the same factor applied to increase the loading rate. This is irrelevant if the material is rate insensitive. If strain rate sensitivity is being modeled, erroneous solutions can result. It is generally desirable to analyze a model in its natural time period if rate dependency is being considered. This can be accomplished through mass scaling.
L5.34
Mass Scaling
As shown in Lecture 1, an estimate of the stability limit in the explicit dynamics procedure can be expressed as
Le t = , cd
where Le is the smallest characteristic element length and cd is the dilatational wave speed of the material. The dilatational wave speed in a linear elastic material (with Poissons ratio equal to zero) is E cd = ,
where E is the elastic modulus and is the material density. If we artificially increase the material density by a factor of f 2: The wave speed decreases by a factor of f. The stable time increment increases by a factor of f.
Copyright 2005 ABAQUS, Inc.
L5.35
Mass Scaling
By artificially increasing the stable time increment through mass scaling, we can analyze the model in its natural time period. Mass scaling has the same influence on inertia effects as artificially increasing the loading rates. Therefore, excessive mass scaling can lead to erroneous solutions. Fixed mass scaling Mass scaling for quasi-static analyses is usually performed on the entire model once at the beginning of the step. One way to define fixed mass scaling is through the specification of a mass scaling factor. Specify f 2 such that the density of every element in the model is increased by f 2. The element stable time increment, thus, increases by f.
L5.36
Mass Scaling
Items affected by mass scaling Mass, rotary inertia, rigid, and infinite elements Rotary inertia in beams and shells Bulk viscosity and mass proportional damping Items not affected by mass scaling Gravity loads Adiabatic heat calculations Thermal solution response in a fully coupled thermal-stress analysis Equation of state materials Fluid and fluid link properties Spring and dashpot elements
L5.37
Mass Scaling
Example: Quasi-static collapse of a curved beam. A curved beam is compressed to 70% of its original length. Shell elements (S4R) with the constitutive behavior of a mild steel are used to model beam.
beam cross-section
L5.38
Mass Scaling
*FIXED MASS SCALING, FACTOR=16
Fixed mass scaling
Scale density of the whole model by a factor of 16 (increases stable time increment by a factor of 4).
L5.39
Mass Scaling
This figure shows the results of four different analyses (contours of PEEQ). The top two results are almost identical. The solution for the results in the second plot requires one-fourth the computer time of the first solution. The last two solutions are essentially meaningless compared to the original static solution.
16
64
Video Clip
6400
L5.40
Mass Scaling
Results ABAQUS results exhibit very good correlation with experimental study
- mass scaling factor 1 - mass scaling factor 16 - mass scaling factor 64
L5.41
Mass Scaling
There are three alternatives to specifying the scaling factor directly:
1 *FIXED MASS SCALING, TYPE=UNIFORM, DT=dt
Scales all elements by a single factor so that the minimum stable time increment of the specified elements becomes equal to dt. Similar to the FACTOR parameter. Applicable to quasi-static analyses.
2 *FIXED MASS SCALING, TYPE=BELOW MIN, DT=dt
Scales only elements whose stable time increment is below the value assigned to DT so that their stable time increment equals dt. Applicable to quasi-static and dynamic analyses.
Copyright 2005 ABAQUS, Inc.
L5.42
Mass Scaling
3 *FIXED MASS SCALING, TYPE=SET EQUAL DT, DT=dt
Scales all specified elements to have the same stable time increment equal to dt. Applicable to quasi-static analyses.
L5.43
Mass Scaling
Variable mass scaling There are situations in which it is desirable to mass scale elements periodically during a step in quasi-static analyses. This might be the case if some elements experience such large deformations that their stable time increment is drastically reduced. The variable mass scaling option provides this capability. Mass scaling calculations are performed periodically during the step. You specify how frequently the mass matrices are updated.
L5.44
Mass Scaling
Variable mass scaling can be used to scale the mass matrix above and beyond that done with fixed mass scaling. Typically variable mass scaling is defined by specifying a target stable time increment. As with fixed mass scaling you can choose to scale all the elements (uniformly or nonuniformly) or only the elements below the target. Alternatively, fully-automatic variable mass scaling is available for bulk metal rolling problems. The mass scaling factor is computed based on mesh geometry and initial conditions and is adjusted throughout the analysis.
L5.45
Mass Scaling
Example: Compression of a jounce bumper Recall from Lecture 1 that the stable time increment for the jounce bumper analysis decreased due to element deformation. Excerpt from status (.sta) file:
STEP TOTAL STABLE INCREMENT TIME ... INCREMENT . . . 5892 3.000E+00 ... 5.087E-04 6876 7861 8876 10702 12890 . . . 3.500E+00 ... 5.074E-04 4.000E+00 ... 5.070E-04 4.500E+00 ... 3.120E-04 5.000E+00 ... 2.739E-04 5.500E+00 ... 2.065E-04 CRITICAL ELEMENT 485 485 485 1247 1247 1247 KINETIC ENERGY 2.457E+02 2.730E+02 3.581E+02 4.106E+02 4.208E+02 3.859E+02
element 1247
element 485
L5.46
Mass Scaling
Example (contd): Compression of a jounce bumper
*VARIABLE MASS SCALING, FREQUENCY=1, DT=0.0005, TYPE=BELOW MIN
Scale density of elements whose stable time increment is less than .0005 s variable mass scaling
L5.47
Mass Scaling
Example (contd): Compression of a jounce bumper Excerpt from status (.sta) file of analysis with mass scaling:
STEP TOTAL STABLE CRITICAL KINETIC PERCENT INCREMENT TIME ... INCREMENT ELEMENT ENERGY CHNG MASS . . . 17846 9.000E+00 ... 5.000E-04 1061 1.184E+01 6.534E-01 18847 9.501E+00 ... 5.000E-04 1108 1.427E+00 6.779E-01 19846 1.000E+01 ... 5.000E-04 1051 4.621E-01 6.816E-01 ODB Field Frame Number 10 of 10 requested intervals THE ANALYSIS HAS COMPLETED SUCCESSFULLY
L5.48
Mass Scaling
Example: Rolling of a thick plate Automatic mass scaling based on mesh geometry and initial conditions.
*VARIABLE MASS SCALING, TYPE=ROLLING, ELSET=METAL, FREQUENCY=1, FEED RATE=.95, EXTRUDED LENGTH=2.0E-3, CROSS SECTION=81
Undeformed shape
L5.49
Mass Scaling
Selected regions of the model can be scaled independently. Useful when different regions of the model have different stiffness and mass properties Example:
*FIXED MASS SCALING, ELSET=plate, FACTOR=1600
Only one fixed and one variable mass scaling factor definition is allowed per element set. An element that has multiple fixed or variable mass scaling definitions results in an error message.
L5.50
Mass Scaling
Mass scaling in multistep analysis Fixed mass scaling definitions are not retained from step to step, but the scaled mass matrix is retained. Using the FIXED MASS SCALING option with no parameters resets the entire mass matrix.
L5.51
Mass Scaling
Variable mass scaling definitions are retained from step to step. Any variable mass scaling definition in the current step causes all variable mass scaling definitions from previous steps to be removed. To remove variable mass scaling from previous steps, use the *VARIABLE MASS SCALING option with no parameters.
Deactivating variable mass scaling does not prevent scaled masses from being carried forward to subsequent steps. To return to the initial mass matrix, the mass must also be reinitialized (i.e., include the *FIXED MASS SCALING option with no parameters).
Copyright 2005 ABAQUS, Inc.
L5.52
Mass Scaling
Avoid very large changes in element mass due to mass scaling at the start of a step that is not the first step in a multiple step analysis. The resulting precision problems in the mass calculations may give rise to erroneous or misleading results. If a large change is desired, do the following:
1 Insert a new step to reinitialize the element masses to their original
values.
2 In a subsequent step add mass scaling definitions to scale the
L5.53
Mass Scaling
Output variables The following mass scaling output variables are available: EMSF EDT DT DMASS Element mass scaling factor. Element stable time increment. Time increment Percent change in mass caused by mass scaling.
ABAQUS/Viewer can be used to create history plots of all four output variables; in addition, contour plots of EMSF and EDT can be created. The time increment and percent change in mass are also printed to the status (.sta) file.
Copyright 2005 ABAQUS, Inc.
Mass scaling factor for jounce bumper with variable mass scaling
Viscous Pressure
L5.55
Viscous Pressure
Viscous pressure (VP) loading is a very effective way to damp out dynamic effects quickly, and thus reach quasi-static equilibrium in a minimal number of increments. It is commonly used to damp out kinetic energy associated with structural motion (usually on the exterior surface of a body). The viscous pressure load is defined through:
p = cv v n.
The effect of the load is that pressure waves crossing the free surface are absorbed; there is no reflection of energy back into the interior of the model. Note that viscous pressure applies damping at the surface of the body. Damping in the interior of the body is introduced through material damping.
L5.56
Viscous Pressure
The choice of viscous pressure coefficient (cv) is critical for using the technique effectively. The value of cv is problem dependent. Typically cv is set equal to a small percentage (1 or 2%) of cd. Usage:
*DLOAD
element_set, VPn, cv
or
*DSLOAD
surface_name, VP, cv
L5.57
Viscous Pressure
Example: Crimp forming A variation on the crimp forming problem discussed previously (Lecture 4, Contact Modeling) As before, the rigid punch undergoes a downward stroke to complete the crimp forming. The anvil is held fixed.
Grip (36 mm thick) Punch Wires (31 mm dia)
Anvil
L5.58
Viscous Pressure
Even though the intended simulation is quasi-static, an explicit dynamic simulation is used. The following aspects would present difficulties for a static analysis with ABAQUS/Standard: The model has no static stability due to the free rigid body motion of the grip and wires. During crimping the grip arms buckle as they are turned by the punch downward into the bundle. There is complex multi-body contact in the analysis: between the grip arms and the wires, between each combination of two wires, and between the two grip arms. The punch is moved downward in such a way as to conduct the analysis efficiently without having inertia effects significantly influence the solution. Viscous pressure is used to damp out dynamic effects quickly.
L5.59
Viscous Pressure
*Heading : *Material, Name=copper *Density 8.5e-3, *Elastic All exterior 17.8e+03,0.34 faces of wires : *Step *Dynamic,explicit ,.2 ** ** Punch Velocity of ~ 36 M/sec ** *Boundary,amplitude=ramp PunchRef,2,2,-7.163 ** ** Viscous Pressure on Wires ** *Dsload,amplitude=ramp4 wires, vp, 0.12 : *End step
cd =
E (1 ) 15 (1 + )(1 2 )
cv / cd 0.8%
L5.60
Viscous Pressure
Deformed configuration
crimping
L5.61
Viscous Pressure
Comparison of viscous and internal energy The plot reveals that ALLKE and ALLVD stay a very low fraction of the ALLIE in the stamping process, so this is indeed a quasi-static problem.
ALLIE
ALLKE, ALLVD
L5.62
Viscous Pressure
Experimental verification The solution is verified by experimental results.
Summary
L5.64
Summary
Excessive loading rates can produce solutions with significant inertia effects. A general guideline is to restrict loading rates to less than 1% of the material wave speed. Ramping applied loads and boundary conditions from zero also promotes a quasi-static response. Use the SMOOTH STEP amplitude definition. Mass scaling can be used to treat rate-dependent material behavior, allowing the process to be modeled in its natural time period.
L5.65
Summary
The energy balance can be used to assist in evaluating whether a given solution represents a quasi-static response to applied loads. Since results can depend strongly on the process speed (real or artificially adjusted by mass scaling), it is vital to ensure that unrealistic results are not being generated by excessive artificial process speed scaling. To confirm that the ABAQUS/Explicit results are realistic, it may be useful to study a simplified version of the problem as a static analysis in ABAQUS/Standard for comparison. The easiest way to create a suitable simplified test case for this purpose is often to define a two-dimensional version of part of the problem.